Summary Bullets:

• Critical Digital Shift: Hyperscalers and data center providers have started competing for cloud networking services, competing for enterprise dollars in space previously dominated by telcos.

• New AI Driven Dynamic: The emergence of enterprise AI is highlighting the importance of network infrastructure and the need to run more distributed workloads, opening new ways for telcos and hyperscalers to collaborate.

Generally telcos and hyperscalers are cautious collaborators. After a brief period where telcos tried to use their pre-existing data center assets to compete in the emerging cloud market, most have decided to move on from those assets. With the emergence of hyperscale data centers, their ubiquitous presence and common operating platforms, they have slowly taken over the market. While some telcos still offer private data center services, most have given up data center assets and instead moved their own IT environments to hyperscale platforms. Some enterprise focused telcos will still function as cloud service providers, migrating applications and maintain the underlying infrastructure they, for the most part, do not own the data centers themselves. While this may be some lost revenue for telecoms, this area was never their specialty.

For the telcos, the emerging threat now is that some of the hyperscale platforms are offering their own networking services. SD-WAN and SASE has transformed how enterprises architect their networks. With the optimization, acceleration, and security tools available, enterprises no longer need to purchase dedicated WAN services and can instead go over the top using internet. This has opened up the WAN market to over the top providers (e.g., Megaport) to sell enterprise networking services that can intelligently route to different cloud environments just using internet services, meaning owning the network and offering private connectivity (which is more expensive) is relatively less valuable. This has spurred the emerging Network-as-a-Service (NaaS) market. Telcos are now launching their own NaaS, but they are not mature offerings. Despite this there is still reliance on carrier networks to provide underlying infrastructure, especially in the last mile.

The rapid increase in the use of cloud and now AI platforms can be a blessing for telcos. Traffic to cloud environments continues to increase. More and more enterprise data is stored in cloud environments, and increasingly large enterprises are using more than one cloud provider. Telcos are now trying to take advantage of this by investing in more infrastructure to connect cloud environments to enterprise locations. The use of AI and particularly Gen-AI required massive data transfer if AI models are in a different physical location than the data that powers them. This is part of the impetus behind renewed telco focus on fixed infrastructure, for example Telstra’s investment in the inter-city fiber project and strategic partnership with Microsoft.

But, hyperscalers as well as data center providers like Equinix are investing in their own networking services, as cloud and networking services are becoming increasingly interlinked. Performance of applications depends not only on the underlying data center infrastructure but also the network that connects it to an end user. Companies including Google and Microsoft are investing in undersea cable routes to have better control over how they route traffic to their services. Meanwhile Equinix is building a software defined platform that connects its global network of enterprise and hyperscale data centers. However, in specific markets, the terrestrial in-ground transport networks, the metro networks and the last mile networks connecting consumers and businesses to data centers and clouds are still controlled by the telcos. While in some cases these may be dumb pipes, there are some advantages telcos still have.

With the advent of AI, the amount of data used, the sensitivity of data used, and the latency required for certain applications, is actually increasing interest in running enterprise applications in more local locations. The telcos still have more distributed networks and have an advantage when connecting distributed workloads. As the enterprise AI market matures we will see a new era of simultaneous collaboration and competition between the hyperscale and telco industry for the provision of digital infrastructure.